Google has launched Gemini 3.7 Flash, bringing improved coding, reasoning, web development and AI agent capabilities. The new model also costs half as much as Gemini 3.6 Flash, making it a more affordable option for developers and businesses.
Google has introduced Gemini 3.7 Flash, its latest AI model focused on coding, knowledge-intensive tasks and AI agents.
The new model arrives just weeks after the launch of Gemini 3.6 Flash and is designed to deliver stronger performance while reducing the cost of running AI-powered workflows.
Google says Gemini 3.7 Flash brings improvements across software engineering, web development, knowledge work, automation and multi-step agentic tasks.
Gemini 3.6 Flash was introduced in July 2026 as a more efficient model for coding, knowledge work and multimodal tasks.
It was designed to reduce token usage and the number of reasoning steps and tool calls required for multi-step workflows. Google priced it at $1.50 per million input tokens and $7.50 per million output tokens.
Gemini 3.7 Flash takes that approach further, with Google highlighting better accuracy, stronger reasoning and improved agent performance.
| Gemini 3.6 Flash vs Gemini 3.7 Flash | ||
|---|---|---|
| Feature | Gemini 3.6 Flash | Gemini 3.7 Flash |
| Focus | Coding, knowledge work and multimodal tasks | Coding, AI agents, web development and complex workflows |
| Coding | Improved coding accuracy and fewer unnecessary edits | Stronger debugging, issue resolution and production-ready code |
| Web development | Improved agentic development | More functional layouts and complete apps with fewer prompts. |
| UI generation | Strong multimodal capabilities | Better adherence to screenshots, images and design systems |
| Reasoning | More efficient multi-step reasoning | More deliberate planning and stronger tool use |
| Developer experience | Reduced retries and execution loops | Better instruction following and intent clarification. |
| API price | $1.50/1M input; $7.50/1M output | $0.75/1M input; $3.75/1M output during introductory period |
The biggest difference is therefore not simply speed. Google is positioning Gemini 3.7 Flash as a model that can think through complicated tasks more carefully, use tools more effectively and complete workflows with less human intervention.
Coding is one of the major areas where Google says Gemini 3.7 Flash has improved over its predecessor.
The model reportedly performs better at tasks such as debugging and resolving software issues. It also delivers higher first-pass code accuracy and is designed to produce more production-ready code.
Google points to improvements in evaluations including FrontierCode 1.1 Main and DeepSWE v1.1, suggesting that developers may need fewer retries and manual corrections when using the model for software engineering.
Gemini 3.7 Flash is also designed to make AI-assisted web development more capable.
Google says the model can generate more functional layouts and build feature-complete applications with fewer prompts. Its UI generation capabilities have also been improved, particularly when developers provide a reference such as a screenshot, an image, or a complete design system.
On Arena.ai's WebDev Arena, Gemini 3.7 Flash reportedly achieved an Elo score of 1,588, compared with 1,538 for Gemini 3.6 Flash.
The improvements are not limited to coding. Google says Gemini 3.7 Flash performs better in knowledge-heavy areas such as finance, law and biosciences.
The model has also shown stronger results on the GDPval benchmark, which tests how well AI models handle complex professional documents.
Google says it also outperforms Gemini 3.6 Flash on AutomationBench, demonstrating better performance in real-world business workflows.
Another major focus of Gemini 3.7 Flash is agentic AI.
The model is designed to handle multi-step tasks by planning more carefully, deciding when tools are needed and following instructions more precisely.
Google says this can reduce the amount of manual supervision required of developers and the number of retries in complex workflows.
Gemini 3.7 Flash is also being used to improve Gemini Spark, Google's 24/7 personal AI agent. The updated model is expected to improve Spark's ability to work with Google Workspace applications and to complete complex tasks that require multiple skills.
Google is offering Gemini 3.7 Flash at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens through the end of the year.
That makes the new model substantially cheaper than Gemini 3.6 Flash, which launched at $1.50 per million input tokens and $7.50 per million output tokens.
The combination of lower pricing and stronger performance could make Gemini 3.7 Flash particularly attractive for companies looking to deploy AI agents at scale.
Developers can access Gemini 3.7 Flash through the Gemini API, Google AI Studio, Google Antigravity and Android Studio. It is also available through Google's enterprise AI offerings.
Google says Gemini Spark is available to Google AI Pro and Ultra subscribers in more than 160 countries, where supported.
With Gemini 3.7 Flash, Google continues to push its Flash series beyond simply providing fast answers.
The focus is increasingly on AI systems that can plan, use tools, write code, and complete complex tasks with less human supervision, while keeping costs low enough for large-scale deployment.
